Recent progress in synthetic biology has established light as a leading tool for observation as well as stimulation of synthetic constructs in cellular contexts (MacDonald and Deans, 2016; Toettcher et al., 2011a; Deisseroth, 2011; Müller et al., 2015; Krueger et al., 2019; Baumschlager and Khammash, 2021; Gheorghiu et al., 2021; Forlani and Di Ventura, 2021; Pérez et al., 2021). Light can be precisely localized in space and instantaneously turned on and off. Achieving this degree of spatio-temporal control establishes light as an exceptionally attractive alternative to other existing stimulation techniques such as chemical induction (Fracassi et al., 2016). Inspired by photosensitive proteins found in nature, synthetic biologists have engineered phototosensors and photoreceptors into living cells to control cellular processes such as gene expression (Müller et al., 2015) (Figure 1) and protein-protein interaction (Toettcher et al., 2011b). This use of light to influence cellular processes is commonly referred to as optogenetics (Toettcher et al., 2011a; Deisseroth, 2011). Several optogenetic tools, pertaining to different cell types and spanning different wavelengths of light, have been developed and studied in the last decade (Tischer and Weiner, 2014; Müller et al., 2015; Repina et al., 2017; Kolar et al., 2018; Baumschlager and Khammash, 2021; Gheorghiu et al., 2021; Forlani and Di Ventura, 2021; Pérez et al., 2021). Applications of these optogenetic tools have been expanding rapidly in various disciplines, such as developmental biology (Johnson and Toettcher, 2018; Krueger et al., 2019).

The ability to provide accurate, tunable, and controlled delivery of light is crucial for the effective application of optogenetics. In recent years, there has been a rapid acceleration in the development of new strategies and hardware-software platforms fulfilling this requirement (Figure 1). Custom-designed light stimulation devices (optogenetic platforms) targeted at different cell culture volumes are now available to address the specific needs of various applications. Certain specialized devices can further facilitate light induction capability at intracellular resolution and support high-throughput operation. Many of these platforms also include dedicated devices to measure cellular outputs such as fluorescent protein reporters (Rodriguez et al., 2017) (Figure 1) or cell growth, enabling real-time observation of target cell behaviour in response to light stimulation.

Optogenetic platforms equipped with a combination of light stimulation and cellular activity measurement devices have allowed for the implementation of in silico feedback control (Figure 2) of cellular processes ranging in application from simple gene expression regulation (Kumar et al., 2021) to multicellular morphogenesis studies (Hartmann et al., 2020). An in silico feedback control loop first involves the quantitative measurement of a desired cellular output (e.g. fluorescence) by a suitable measurement device. The output measurement data are then sent to a control computer which performs dedicated computations over the data using feedback controllers (Figure 2) such as integral and model predictive control strategies to determine a light intensity that the target cells should be stimulated with in order to achieve a desired output behaviour (e.g. set-point tracking). The computed light intensity is then applied to the target cells via the light stimulation device. This sequence of measurement, computation, and stimulation are executed iteratively at fixed time intervals to achieve closed-loop feedback control of the target process.

This review aims to introduce and describe various optogenetic platforms and in silico control strategies developed for synthetic biology studies. In this article, we first categorize these platforms into three broad groups based on their intended target cell culture volume ( < 1 m l – low; 1–100 ml–medium; > 100 m l – large), and discuss their design methodologies in their respective applications, before offering our perspective on their further development. We further review a few unique platform designs intended for emerging optogenetic applications such as in animal implants and 3-D spatial induction. Finally, we conclude with a brief discussion on existing in silico feedback control strategies and applications. For brevity, the main focus is on optogenetic platforms demonstrated for non-neuronal cell studies, and platforms dedicated for neuronal (Warden et al., 2014; Kim et al., 2017; Vázquez-Guardado et al., 2020) and plant (Ochoa-Fernandez et al., 2020) cell studies are not discussed.

This category spans optogenetic platforms with target cell culture volumes ranging from a few microliters to a milliliter. These platforms are usually employed for short-term experiments to characterize and optimize optogenetic tools and downstream genetic circuits. They are also effective in studies such as pattern formation, requiring light-stimulation and observation at intra-/inter-cellular resolution. These platforms can be further divided into two groups: set-ups having a light stimulation device coupled with microscope systems, and set-ups having a dedicated LED-array light induction device adapted for micro-multiwell plates. The first group of platforms mostly facilitates higher spatial resolution and dynamic studies, whereas the second group is usually considered in high-throughput experiments pertaining to steady-state studies.

This section spans optogenetic platforms with intended target cell culture volumes of a few to hundreds of milliliters. These volumes are usually considered in studies where certain constraints on growth conditions or cell density are needed during long-term experiments. For example, in the optogenetic experiments in (Aditya et al., 2021), cells needed to be continuously maintained in exponential growth phase. Low-volume culture platforms are not suitable for those studies as it is difficult to regulate cell density at such low volumes. Most of the platforms described here are coupled with turbidostat, chemostat, or morbidostat set-ups in order to facilitate operation in those conditions. A generic architecture of these platforms is shown in Figure 3C.

One of the initial designs in this category was proposed by Milias-Argeitis et al., 2011 in their pioneering work of demonstrating in silico feedback control of gene expression in yeast. This set-up involved a custom-built light delivery system with a light pulser device (having two triplets, red and far-red wavelengths, of high power LEDs) and an electronics box containing LED driver electronics which can be connected with a computer via a multifunction DAQ (Data Acquisition) board. The custom-designed electronics provided both manual as well as computer-controlled (needed for feedback control experiments) operation of the device. This system can be used both with multiwell cell culture plates (without independent illumination of separate wells) by placing sample plates underneath the light pulser device (Milias-Argeitis et al., 2011), and with glass/transparent tube cultures (Ruess et al., 2015). The authors in (Ruess et al., 2015) used this system with a hot plate magnetic stirrer to maintain a desired cell culture temperature and provide sufficient aeration to the target culture. They also diluted the culture manually at regular intervals to keep the cell density within reasonable bounds during an experiment. Another simple design with a higher-throughput set-up (capable of running 15 tubes independently in parallel) was used by Baumschlager et al., 2017 (also in (Benzinger and Khammash, 2018)). This set-up comprises a water bath (to maintain a desired temperature) having custom-built 15-tube holders, which can hold 25-ml glass cell culture tubes, each fitted with a magnetic stirring device and a custom-made LED pad (with a white diffusion filter to ensure homogeneous illumination) located underneath each culture tube. This set-up allows for having a pre-programmed illumination sequence for each tube in an experiment, but does not allow for the real-time change of LED intensity during the experiment, and thus doesn’t have feedback control operation capability. A similar but improved Light Tube Array (LTA) platform was proposed by Olson et al., 2014 which is capable of providing an isolated optical environment for 64 standard 14-ml culture tubes. Each tube can be illuminated with four different LEDs independently with the given LED driver electronics that can be pre-programmed with light-intensity sequences. Although the set-up does not include any stirring or temperature control mechanism in itself, the whole tube rack can be placed in a suitably-sized shaking incubator.

Several other innovative and advanced platforms combining multiple functionalities and automation principles have been proposed as well as demonstrated in various applications, notably feedback control over cellular processes (owing to the availability of both light-stimulation and measurement devices on these platforms). In (Melendez et al., 2014) and improved in (Stewart and McClean, 2017), the authors exhibited controlling intracellular protein concentration with a CRY2-CIB1 optogenetic construct in yeast by using their automated culturing optogenetic platform. The set-up they proposed involves a 27-ml working volume transparent culture vessel where cells can either be grown in batch culture or maintained as continuous culture for several days in chemostat mode by using the automated media influx and culture efflux system. Provisions for temperature control, agitation and aeration to oxygenate cells in the vessel are available on this platform. A separate microscope-based measurement apparatus is also integrated with the set-up, allowing fluorescence imaging and quantification of single cells in the culture by running them through a single channel microfluidic flow cell mounted on the microscope. Light induction in the culture vessel is performed with a high-powered blue LED, all controlled by a computer via a custom-designed control board. Another automated optogenetic platform demonstrated by Milias-Argeitis et al., 2016 employs a custom-designed turbidostat to maintain cell culture at a desired optical density (OD), and an automated sampler to facilitate real-time measurement via flow cytometer. This platform also has integrated culture incubation, LED-based light induction, and a culture stirring system for proper aeration in the 25-ml glass centrifuge cell-culture tube. By using computer control over the whole apparatus, Milias-Argeitis et al., 2016 exhibited optogenetic cell growth-rate control of a culture of bacterial cells for over 24 h. Another similar platform was proposed by Harrigan et al., 2018 which comprises eight (in two incubator arrays of four each) or more (expandable) separate temperature-controlled and magnetically-stirred chambers for 50-ml conical cell culture tubes and LED stimulation. The authors also designed an automated integrated liquid handling module capable of moving cell culture from individual chambers to a flow cytometer for measurement or to a waste reservoir for dilution as well as replenishing the individual chambers with fresh media.

Recently, researchers have put forth various innovative and compact optogenetic platform designs pertaining to this medium-volume culture category. One particularly unique design, Chi.Bio, was presented by Steel et al., 2020 which establishes a very compact and low-cost approach of optogenetic stimulation and in situ fluorescence measurements. Their system is divided into reactors, pump boards and mini control computers. The reactor is a compact housing for 30-ml clear glass test tubes (open to the atmosphere), and it has integrated tunable light-induction LEDs, temperature control and stirring systems, an OD sensor, and a spectrometer for measuring multiple fluorescence proteins in situ. One pump board per reactor is needed to achieve media or culture flow to/from the reactor culture tube providing turbidostat/chemostat/morbidostat operation capability. Furthermore, the mini control computer can control eight such integrated reactor pump board set-ups independently in parallel. The software stack enables users to control and monitor any peripherals, and set-up experiments via a user-friendly web user interface. Any advanced procedures and computations in an automated experiment, for example feedback control algorithms, are easily implementable via their python framework. One downside of this platform is the low sensitivity of its low-cost integrated chip-based spectrometer compared to much more expensive flow-cytometers. This might be a limiting factor in some applications with synthetic genetic constructs requiring very sensitive fluorescence measurements. But one can also integrate separate liquid handling modules with this set-up to facilitate flow-cytometric or microscopic measurements in addition to the in situ fluorescence measurement. The ReacSight strategy proposed by Bertaux et al., 2021a offers such an enhancement. This is a generic and flexible strategy for adding different instruments in the automated measurement framework of different optogenetic platforms. Bertaux et al., 2021a also demonstrated this strategy with Chi.Bio reactors as an example, and this combined platform of ReacSight and Chi.Bio has been further used in some recent optogenetic control studies (Aditya et al., 2021, 2022).

Another interesting optogenetic platform with a custom-designed multicolor fluorescence monitoring device was proposed by Pen et al., 2021. Here, the cell culture chamber is a continuous cell culturing module (allowing turbidostat or chemostat mode operation) adapted for a 30-ml cylindrical glass vial with a gasket cap (for gas handling functions), and includes temperature control and magnetic stirring systems. Additionally, this set-up by Pen et al., 2021 also has a separate module for in situ fluorescence measurement. There is a provision of dipping a multifiber (optic bundle) probe into the cell culture through the lid which allows excitation light propagation from excitation LEDs onto the cell culture and subsequent collection of fluorescence emission signals which are then guided towards a microPMT (Photomultiplier tubes) for fluorescence measurements. These PMTs are expensive but have a higher sensitivity and higher dynamic range compared to the fluorescence sensor used in Chi.Bio. This platform can also be modified and used for light stimulation of the cell culture via the fluorescence excitation LEDs.

Similar to the LED-array light induction devices discussed in the previous section, these optogentic platforms are only intended for target stimulation and observation on a population level, meaning they cannot be employed for single-cell studies. Standalone optogenetic platforms without a measurement device can cost around $10k or less for 10–16 parallel throughput set-ups. However, to achieve feedback control, one would need to integrate them with other measurement devices like flow-cytometer or microscope which are very expensive. As mentioned previously, the Chi.Bio set-up (Steel et al., 2020) does offer a low-cost integrated spectrometer, but it is not suitable for measuring weak fluorescence signals. Unlike low-volume culture platforms, most of these devices have integrated turbidostat, chemostat, or morbidostat operation set-ups which can be used to maintain growth conditions and regulate cell densities in the target culture during an optogenetic experiment. Moreover, most of these platforms can only work with suspension cell cultures. In the non-optogenetic biology literature, there has been extensive development of turbidostat/chemostat/morbidostat set-ups; therefore, a wide variety of designs, intended for different applications, are now available (Toprak et al., 2013; Miller et al., 2013; Takahashi et al., 2015; Hoffmann et al., 2017; Wong et al., 2018; Bergenholm et al., 2019; McGeachy et al., 2019; Gopalakrishnan et al., 2020). With some basic electronics know-how, one can easily integrate light-stimulation LEDs or other light sources into these already existing devices to achieve optogenetic stimulation.

Here we consider optogenetic platforms aimed at suspension cell culture volumes beyond hundreds of milliliters. These large volumes are usually handled in large-scale bioreactors which are one of the staple devices in biochemical/bioprocess industries (Najafpour, 2015). Given the scope of this review article, we will restrict our discussion to bioreactor platforms employed in academic research.

There have been a number of incremental studies on metabolic engineering and protein production regulation in lab-scale bioreactors. One of the major goals of metabolic engineering research today is to maximize the yield of a given bioproduction process. To achieve this goal, there is an ever-growing interest in the synthetic biology community to build genetic constructs allowing for dynamic control of metabolism (Venayak et al., 2015; Lynch, 2016; Ni et al., 2021). The idea is to introduce externally inducible systems in metabolic pathways with different induction strategies (Lalwani et al., 2018; Bertaux et al., 2021b). Several induction strategies (e.g. chemical induction, temperature control etc.) have been explored in this context. A list of these strategies, pertaining to Escherichia coli and Saccharomyces cerevisiae cell-cultures, has been summarized in (Lalwani et al., 2018). With the recent development in optogenetics and related genetic tools, one can not help but foresee its application in metabolic engineering with its unique benefits (light induction is inexpensive and provides precise temporal control) (Carrasco-López et al., 2020; Pouzet et al., 2020). However, this strategy of using light induction for efficient bioproduction is still in its nascent phase of realization, and only a few studies with simple light delivery devices have been carried out in recent years.

In (Zhao et al., 2018), the authors demonstrated production of isobutanol and methyl-1-butanol via light stimulation of an engineered S. cerevisiae culture. In their study, they used 500-ml, 2-L and 5-L capacity bioreactors with custom-designed light delivery set-ups. Multiple (2 for 500-ml and three for 2-L) LED matrix panels were used to surround the first two (500-ml and 2-L) transparent glass bioreactors, while multiple blue LED strips were wrapped around the 5-L bioreactor to achieve light stimulation of respective cell-cultures. A similar set-up for a 2-L bioreactor was used by Lalwani et al., 2021 to demonstrate light-induced mevalonate and isobutanol production in engineered E. coli. In (Zhao et al., 2021), the authors presented a series of sensitive optogenetic circuits (OptoAMP) and demonstrated their ability to enhance the production of lactic acid, isobutanol, and naringenin on a photo-bioreactor platform. They used a commercial high power LED-matrix panel to illuminate the 5-L bioreactor in their experiments. These simple light delivery devices are inexpensive and can be coupled with other suitable (transparent) bioreactors which are usually expensive on their own. A representative design of these platforms is illustrated in Figure 3D.

One problem encountered when illuminating a large volume culture is insufficient penetration of light at high cell densities (Carrasco-López et al., 2020). This challenge can be overcome by either improving the sensitivity of optogenetic circuits (Zhao et al., 2021), risking a reduction in dynamic range and an increase in leaky dark activation, or by improving the light delivery system. One can design the inside of a bioreactor with provisions for additional light sources to enhance light penetration, for example, having multiple LED rods dipped in the target cell culture via suitably designed lids. Similar design strategies have been proposed by Carrasco-López et al., 2020 in their opinion article, but these designs have not been implemented yet. Within the realm of microalgae biotechnology, large photobioreactors are commonly used for microalgae cultivation. One can also employ their designs (Posten, 2009; Chen et al., 2011) for bioreactor illumination with little or no modifications. Furthermore, to provide feedback control, a liquid handling module similar to the design proposed by Harrigan et al., 2018 can be used to move samples from the bioreactor to a suitable measurement device such as flow-cytometer. To the best of our knowledge, no optogenetic feedback control as envisioned in (Carrasco-López et al., 2020; Pouzet et al., 2020) has been demonstrated yet with these bioreactor platforms.

Apart from the devices and systems covered in the previously discussed three categories (low-, medium-, and high-volume culture platforms), there are some other innovative and unique designs which were developed for specific applications. We would like to mention a few of these platforms which may have potential utility in emerging optogenetic applications.

A compact light delivery device (LED Thermo Flow) that can be directly connected to flow-cytometer injection ports was described by Brenker et al., 2016. In this design, a glass FACS tube containing target cell culture is inserted into a custom-built cylindrical sleeve. The sleeve contains multiple wavelength LEDs with manually-controllable intensities surrounded by water which is constantly circulated via a heated water pump to maintain the sample temperature at a desired value. The FACS tube, together with the sleeve, is then connected to a flow-cytometer injection port for direct measurement. This allows one to measure real-time effects of optogenetic stimulation and deduce kinetics of optogenetic tools.

In all the optogenetic platforms described previously, light illumination was targeted for cells growing in any cell culture lab-platform but not for cells proliferating within a live animal. In (Ye et al., 2011), the authors devised a system to provide light-stimulation to mammalian cells (engineered with a desired optogenetic construct) implanted intraperitoneally in mice. In their set-up, light from a high-brightness LED is delivered to the cells in hollow-fibre implants via a multimode silica glass optical fibre. The LED is powered by a regulated DC power supply whose output voltage (in turn, the LED light intensity) can be modulated via Labview software running on a computer.

All the platforms discussed so far in this review facilitated either precise spatial 2-D illumination (microscope-coupled platforms) or full 3-D homogeneous illumination with no spatial control. One of the emerging light induction techniques involves stimulating samples in 3-D with spatial precision. Holography-inspired multiphoton photo-excitation (3D-SHOT) (Pégard et al., 2017), DMD-based patterned illumination in lightsheet microscopy (Jiao et al., 2018), and three-dimensional multi-site random access photostimulation (3D-MAP) (Xue et al., 2022) are some of the recently proposed designs in this direction. These platforms have the potential to aid artificial tissue engineering or organoid development via optogenetic induction. To the best of our knowledge, these new techniques have only been applied in neuroscience research such as whole-brain neuronal activity studies in larval zebrafish (Jiao et al., 2018). Their application with non-neuronal cells or tissues is yet to be explored.

Feedback control is a key concept in the control theory literature. It has a wide range of applications in our day to day life, from a thermostat controlling room temperature to the autopilot maintaining airplane attitude. Feedback control over a process requires a mechanism to tweak the process as well as a pathway to measure the process behaviour. Here, we will briefly discuss different feedback control methodologies pertaining to cellular biology contexts where the tweaking/stimulation of a target process (such as gene expression) is enabled by optogenetics. A brief introduction and description of optogenetic in silico feedback control procedure has been laid out previously in the introduction section, and a related illustration is shown in Figure 2.

Depending on the application as well as the capability of the optogenetic platform used, one can implement in silico feedback control over a target cell culture in two different settings: population control (Milias-Argeitis et al., 2011; Melendez et al., 2014; Milias-Argeitis et al., 2016; Harrigan et al., 2018) and single-cell control (Rullan et al., 2018; Kumar et al., 2021; Fox et al., 2022), as illustrated in Figure 4. Population control involves an aggregate measurement of the target cell culture and stimulation with a common input light intensity for all the cells in every measurement-computation-stimulation cycle. This can be achieved with those optogenetic platforms which are integrated with a measurement device (assuming that the required controller computation over measurements to compute the input light intensity can be performed on a computer). Meanwhile, exhibiting a reduced cell-to-cell output variability compared to population control (Rullan et al., 2018), single-cell control involves measuring single-cell outputs in a culture, running controller computations for each cell based on its quantified output, and then applying the computed light intensities to the respective target cells independently. This setting can be employed with microscope-coupled projector-based optogenetic platforms discussed previously.

Another critical aspect of in silico feedback control application is the type of controller algorithms employed in the computation. This controller algorithm has the task of computing an input light-intensity (which the target cell culture should be stimulated with) based on output measurements obtained from the target cell culture (in every measurement-computation-stimulation cycle), in order to achieve a desired output behaviour (Figure 2), for example, driving nascent RNA count to a desired set-point value (Kumar et al., 2021). One of the initial demonstrations of optogenetic in silico feedback control was realized by Milias-Argeitis et al., 2011. With their custom-designed optogenetic platform, the authors demonstrated robust regulation (desired set-point tracking) of gene expression fold change in genetically engineered S. cerevisiae by using an in silico feedback controller implementing Model Predictive Control (MPC) strategy. Further, Milias-Argeitis et al., 2016 carried out optogenetic in silico feedback regulation of gene expression as well as cell growth in suitably engineered E. coli cultures. They implemented two different control algorithms: Proportional-Integral (PI) and MPC. PI controllers have a very simple implementation offering a control input (light-intensity) value which is proportional to the sum of 1) the error between a desired set-point and the current measured output, and 2) the time-integral of this error. These PI controllers guarantee zero error at steady-state when the set-point is constant over time, but they are not suitable for tracking time-varying set-points. On the other hand, MPCs are capable of tracking dynamic set-points but they require a model of the target process in their implementation. Further technical details and a comparison between these two controllers are described in (Milias-Argeitis et al., 2016). Another approach using a Proportional-Integral-Derivative (PID) controller has been proposed by Soffer et al., 2021 where the authors employed a model-based design for tuning controller parameters. Other in silico controller implementations such as PI control (Toettcher et al., 2011c), bang-bang control (Melendez et al., 2014), MPC (Harrigan et al., 2018), and integral feedback (Rullan et al., 2018) have been demonstrated with their respective optogenetic platforms and target processes. Interested readers can refer to (Grosenick et al., 2015) for a list and brief descriptions of different controller types which can be realized in in silico control architecture for controlling any suitable target cellular process. We would like to note here that this list is formulated in the context of neuroscience applications, but similar conclusions can be extended to non-neuronal cellular applications as well. A similar list of feedback controllers (including their advantages and disadvantages) along with some examples in non-neuronal studies can be found in (Carrasco-López et al., 2020).

Optogenetic in silico feedback control has numerous emerging applications in biotechnology, biomedicine, bioproduction and various synthetic biology studies (Chen et al., 2013; Liu et al., 2018; Lugagne and Dunlop, 2019; Hartmann et al., 2020; Banderas et al., 2020; Carrasco-López et al., 2020; Pouzet et al., 2020; Perrino et al., 2021; Ruolo et al., 2021). Formation of emerging patterns from lateral inhibition via an optogenetic in silico feedback approach to emulate cell-to-cell communication (Figure 5A) was demonstrated by Perkins et al., 2020 using the optogenetic platform described in Rullan et al., 2018. Furthermore, Kumar et al., 2021 implemented several in silico biomolecular controller motifs in stochastic setting and exhibited transcription regulation in suitably engineered S. cerevisiae via single-cell optogenetics control. With this in vivo (transcription in yeast) - in silico (stochastic controller) hybrid closed-loop set-up, called Cyberloop (Figure 5B), they showed characterization and rapid-prototyping of autocatalytic integral (Briat et al., 2016a), antithetic integral (Briat et al., 2016b), antithetic integral rein (Gupta and Khammash, 2019), and biomolecular PID (Filo et al., 2022) controller motifs. These characterization results provide useful insights for optimal controller performance, which can further guide in vivo implementations of these biomolecular controller motifs.

With the advent and ongoing development of various synthetic optogenetic tools that can be implemented in various cell types, a number of hardware-software platform designs have been co-developed. As discussed in this review, these optogenetic platforms have varying features and design methodologies depending on the application, target cell type, and target cell culture volumes. Their complexities range from simple LED strips to full-fledged microscope-coupled 3D hologram generators. Various simple set-ups, mentioned in this review, comprise readily available low-cost components, and can be easily built/assembled in basic research labs equipped with elementary electronic tools and minimal mechanical design expertise. They have the potential to reduce barriers for carrying out advanced optogenetics research with minimal resources. When integrated with an appropriate measurement device, some of these platforms also provide optogenetic in silico feedback control (Cybergenetics) capability, with which different feedback control strategies can be implemented and investigated. This capability has a wide scope of applications in biotechnology and synthetic biology studies.